JOURNAL OF CHEMICAL PHYSICS

VOLUME 115, NUMBER 3

15 JULY 2001

The A 6 - X 6 transition of CrH, Einstein coefcients, and an improved description of the A state
Charles W. Bauschlicher, Jr.a)
Mail Stop 230-3, Space Technology Division, NASA Ames Research Center, Moffett Field, California 94035

R. S. Ram
Department of Chemistry, University of Arizona, Tucson, Arizona 85721

Peter F. Bernath
Department of Chemistry, University of Waterloo, Waterloo, Ontario, Canada N2L 3G1

C. G. Parsons and D. Galehouseb)

Department of Physics, University of Akron, Akron, Ohio 44325

Received 8 March 2001; accepted 17 April 2001 The spectrum of CrH has been reinvestigated in the 9000 15 000 cm 1 region using the Fourier transform spectrometer of the National Solar Observatory. The 1-0 and 1-1 bands of the transition have been measured and improved spectroscopic constants have been A 6 -X 6 determined. A value for the 2-0 band origin has been obtained from the band head using estimated spectroscopic constants. These data provide a set of much improved equilibrium vibrational and state. An accurate description of the A-X transition has been rotational constants for the A 6 obtained using a multireference conguration interaction approach. The inclusion of both scalar relativity and Cr 3s3p correlation are required to obtain a good description of both states. The ab initio computed Einstein coefcients and radiative lifetimes are reported. 2001 American Institute of Physics. DOI: 10.1063/1.1377892

I. INTRODUCTION

CrH was rst observed in 1937 by Gaydon and Pearse,1 who found a complex band near 368 nm. Since then CrH has been studied by several workers2 6 and a summary of the literature was provided in our previous papers on CrH Ref. 7 and CrD Ref. 8 . In our most recent high resolution studies of the 0-0 band of CrH7 and 0-1, 0-0, 1-0, and 1-2 bands of CrD,8 we obtained much improved molecular constants and X 6 states of CrH and CrD, and from for the A 6 the observed perturbations we have identied a low-lying state near 11 186 cm 1. These observations have been a4 conrmed in a recent theoretical study by Dai and Balasubramanian.9 CrH is a molecule of astrophysical importance. It has already been tentatively identied in the spectra of sunspots10 and S-type stars11 and is also found in the spectra of brown dwarfs.12 Brown dwarfs are substellar objects that lie between the coolest oxygen-rich stars M -type and giant planets such as Jupiter. As the surface temperature of an M star drops below approximately 1800 K, metal oxides such as TiO and VO are replaced by metal hydrides such as CrH and FeH. Thus the presence of the 1-0 and 0-0 bands of the transition of CrH are primary markers for A 6 -X 6 L-type brown dwarfs. As the surface temperature drops below approximately 1400 K, metal hydride bands weaken and are replaced by overtone bands of methane and water in the near infrared region in the T-type of brown dwarfs.
a b

An accurate description of the X and A states and the oscillator strengths are required to calculate the molecular opacities used by astronomers to model the spectral energy distribution emitted by brown dwarfs. The ground X 6 state of CrH is relatively well characterized through the far infrared13 and mid infrared14 laser magnetic resonance measurements. Because there was also some experimental data5,8 for the A state, our initial plan was to use ab initio calculations to compute the transition moment. However, this approach showed that the spectroscopic data for the transition are in need of improvement particuA 6 -X 6 larly for the excited vibrational levels of the A 6 state. Thus a joint computational/experimental approach is used. In the present work we have remeasured the spectra of transition of CrH the 1-0 and 1-1 bands of the A 6 -X 6 and have obtained an improved set of constants for the v 1 vibrational level of the A 6 state. The band origin for the 2-0 band has been determined by tting the band head position using extrapolated rotational constants for the v 2 level. In the ground state all of the available data for v 0, 1, and 2 were combined to determine a set of improved vibrational and rotational equilibrium constants. High level ab initio calculations were performed for the X and A states, which are compared with experiment. The calculations also yielded an accurate transition dipole moment.
II. EXPERIMENTAL APPROACH

Electronic mail: bauschli@pegasus.arc.nasa.gov Current address: 15764 Galehouse Road, Doylestown, OH 44230. 1312

The spectrum of CrH in the 9000 15 000 cm 1 region was recorded using the 1-m Fourier transform spectrometer associated with the McMathPierce telescope of the Na 2001 American Institute of Physics

0021-9606/2001/115(3)/1312/7/$18.00

Downloaded 20 Aug 2001 to 129.97.47.16. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp

J. Chem. Phys., Vol. 115, No. 3, 15 July 2001

The A

-X

transition of CrH

1313

tional Solar Observatory at Kitt Peak. The bands were excited using two kinds of sources, a hollow cathode lamp and a microwave discharge lamp. In the hollow cathode experiment the molecules were formed by owing a mixture of 2.8 Torr of Ne and about 100 mTorr of H2 through a Cr hollow cathode lamp, which was operated at 460 V and 765 mA current. The spectrometer was operated with a GG495 red pass lter and midrange photodiode detectors. In this experiment 19 scans were coadded at 0.02 cm 1 resolution in about 2 h of integration. In the microwave-discharge experiment the chromium was introduced as a volatile compound and reacted in the ow tube. Approximately 15% hydrogen mixed with helium carrier gas was admitted into the feed line. A small amount of CrO2Cl2 was vaporized into the mixed stream of gas. Under these conditions, the chromium is reduced preferentially to CrH during excitation. The pressure was adjusted to 2030 Torr with the aid of a needle valve on the pump line. The chromium concentration was adjusted by controlling the chamber temperature or with a controlled bypass. The mixture was carried through a 11 mm i.d. quartz discharge tube within the microwave cavity. The ow rate affects the time allowed for the reaction and also the Doppler linewidth because of the cooling effect of rapid ows. The operating parameters are not particularly sensitive and a strong CrH spectrum was obtained easily. The spectrum was taken at Kitt Peak National Observatory on May 21, 1985. The spectra were measured using a data reduction program called PC-DECOMP and the line positions were obtained by tting a Voigt line shape function to each spectral feature. The hollow cathode spectrum was calibrated using the measurements of Ne atomic lines by Palmer and Engleman15 while the microwave-excited spectra were calibrated by transferring the calibration from the hollow cathode spectra to the microwave-excited spectra using common lines. The absolute accuracy of the wave number scale is expected to be of the order of 0.001 cm 1 while the precision of lines is expected to vary from 0.002 to 0.004 cm 1 depending on their intensity, signal-to-noise ratio, and blending.
III. COMPUTATIONAL METHODS

The X and A states are treated using the multicongurational self-consistent-eld/multireference conguration interaction approach, MCSCF/MRCI. The MRCI calculations use all MCSCF congurations as reference congurations. In many calculations, internal contraction16 IC is used to limit the size of the calculations. The importance of higher excitations is estimated using the multireference analog of the Davidson correction. In most calculations only the Cr 3d and 4s and H 1s electrons are correlated. The calculations including Cr 3s and 3p correlation are denoted 3s3 p. Scalar relativistic effects are included using the one-electron Douglas Kroll DK approach.17 Several choices of active space are considered. In the calculations labeled MCSCF, MRCI, IC-MRCI, DK-MRCI, and IC-MRCI 3s3p the active space in the MCSCF and MRCI calculations includes the Cr 3d, 4s, 4 p , and H 1s orbitals. In these calculations, the MCSCF calculations are actually a complete active space SCF CASSCF calculation

as all distributions of the active electrons in the active orbitals are used. In C 2 v symmetry there are ve active a 1 orbitals and one active orbital in the b 1 , b 2 , and a 2 symmetries, which we denote as 4111 . In the MCSCF calculations, each component of the 3d and 3d orbitals has an occupation of essentially 1.0 electrons, thus this calculation effectively corresponds to all distributions of three electrons in the Cr 3d , 4s, 4 p , and H 1s orbitals. This calculation can describe bonding from both the 7 S and 5 D states of Cr. The dd calculations indicate the Cr d-d correlation has been included by adding a 3d orbital to the active space, yielding a 7222 active space. To reduce the size of the calculation, we restrict the congurations to those with four electrons in the 3d , 3d , 3d , and 3d orbitals. The p indicates that the Cr 4p orbital has been included in the active space, thus yielding a 5221 active space; to prevent the added orbitals from becoming d correlating in character, we include only those congurations with four electrons in the 3d and 3d orbitals. In the calculation labeled IC-MRCI p dd , the active space includes the Cr 3d, 3d , 4s, and 4p orbitals and the hydrogen 1s orbital, i.e., a 7332 active space. There are no restrictions on the congurations, and thus this MCSCF is actually a CASSCF calculation. Three basis sets are used in this work. The set labeled ANO is the averaged atomic natural orbital set18 for Cr, which can be described as (20s15p10d6 f 4g)/ 7s6 p4d3 f 2g , and for hydrogen, is the augmentedcorrelation-consistent polarized-valence triple zeta aug-ccpVTZ set.19,20 The ANO set starts from this set and uncontracts an additional s, p, and d primitive, as well as adding an additional diffuse s, p, and d function to the Cr basis set; the hydrogen basis set is the aug-cc-pV quadruple zeta set. The diffuse Cr functions are even tempered with a of 2.5. The set labeled GEN uses the hydrogen aug-ccpVTZ set, but recontracts the Cr ANO primitive basis set. The s, p, and d primitives are contracted based on an SCF calculation of the 7 S state of a Cr atom. The inner 16 s primitives are contracted to three functions and the outermost four are uncontracted, the inner 10 p primitives are contracted to two functions and the outermost ve are uncontracted, and the inner 5 d primitives are contracted to one function and the outermost ve are uncontracted. The f and g sets are taken from the ANO set. The GEN basis can be described as (20s15p10d6 f 4g)/ 7s7 p6d3 f 2g , and this contraction is suitable for the study of Cr 3s3p correlation. In the preliminary work, the potentials near r e are studied and e is computed using a parabolic function in r and 1/r, which agree with each other to better than 3 cm 1. In the nal phase, complete potentials are computed and the vibrational levels are determined numerically. Using the separations between the lowest three levels, we computed e and e x e . These potentials along with the computed electronic transition moments are used to compute the radiative lifetimes and Einstein A values. The MCSCF/IC-MRCI calculations are performed using 21 MOLPRO while the MCSCF/MRCI calculations are performed using MOLECULE-SWEDEN.22 The DK integrals are

Downloaded 20 Aug 2001 to 129.97.47.16. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp

1314

J. Chem. Phys., Vol. 115, No. 3, 15 July 2001

Bauschlicher et al.

FIG. 1. The 1-0 band head of the A

-X

transition of CrH.

computed using a modied version of the program written by Hess.

IV. ANALYSIS AND DISCUSSION OF EXPERIMENT

The spectrum of the 1-0 band of the A 6 -X 6 transition observed in the microwave-excited spectrum, particularly for low J lines, is stronger than that obtained in the hollow cathode experiment. In part, this is because the microwave discharge source is cooler than the hollow cathode lamp. Both of these spectra, when used simultaneously, were very helpful in identifying the molecular lines belonging to CrH. The line positions of 1-0 and 1-1 bands provided by OConnor2 at lower resolution were also helpful in the line identication. No bands involving the v 2 vibrational level of the A 6 state were observed. A part of the 1-0 band from the microwave-excited spectrum is provided in Fig. 1, where a few lines near the R 1 and R 2 heads have been marked. It is found that the lines of higher energy spin components F 5 and F 6 are much weaker than the lines of the lower energy spin components F 1 and F 2 . In fact only a few lines of the F 5 spin components were identied in our spectra while no lines could be identied in the F 6 spin component. The line identication in the 1-0 band was accomplished using lower state combination differences obtained from the previous analysis of the 0-0 band of CrH7 while the lines in the 1-1 band were identied by comparing the upper state combination differences from the 1-0 and 1-1 bands. The line positions for the 1-0 and 1-1 bands are presented in the Supplementary Material Section of Ref. 23 as E-PAPS document les. The line positions of the 0-0, 1-0, and 1-1 bands were tted simultaneously with the usual N2 Hamiltonian as given by Brown and Milton24 with matrix elements provided by Gordon and Merer.25 A complete listing of matrix elements with some additional parameters such as H, L, H , and L was tabulated in our previous paper on CrH.7 The nal ts were made in two parts. First the hyperne-free pure rotational transitions26 and the hypernefree vibrationrotation line positions calculated using the constants of Lipus et al.14 were included in a t along with transition and all our previous 0-0 lines of the A 6 -X 6 available ground state combination differences. The constants T v , B v , D v , H v , v , D v , v , D v , v , and s v

were varied in the ground state with T 0 xed to a value of 0.0. The parameters H v , v , and s v were not determined for the v 1 and 2 vibrational levels of the ground state. The 1 and 2 vibrational levD v constants were needed for v els and were xed at the values reported by Lipus et al.14 For v 0, additional higher order parameters are also required in order to obtain a satisfactory t. In the second t the 1-0 and 1-1 lines were added and only the v 1 constants were varied. The lines of the F 6 spin components as well as the high J lines of the other spin components that were observed by OConnor2 were used but with reduced weights. The H v , L v , L v , v , D v , and sD v constants were not determined state. It is found for the v 1 vibrational level of the A 6 that v 1 is affected by several perturbations as can be seen from the observed-calculated differences for the lines of the 1-0 and 1-1 bands. The perturbations show no clear patterns and we are unable to draw any conclusion about the nature of the perturbing state or states . In our previous work on CrH, the observed localized perturbations in the 0-0 band led us to identify the a 4 state lying just below the A 6 state. More than one state may be responsible for the extensive perturbations observed in bands with v 1. The molecular constants for the ground and excited states are provided in Table I. The rotational constants obtained for the v 0 and 1 vibrational levels of the A 6 state have been used to derive the equilibrium rotational constants B e and e . Because the term energy position of the v 2 level of the A 6 state is necessary to derive the equilibrium vibrational constants, we have used the 2-0 band head position (14 510.6 cm 1) as reported by Kleman and Uhler4 to determine the band origin of the 2-0 band. We have xed the extrapolated rotational constants for the v 2 level and the known rotational constants for the v 0 level in our tting program and changed the band origin in such a way that the calculated band head position matches the observed band head position. The 2-0 band origin of T 2 14 468.6 cm 1 obtained in this manner is expected to have a precision of about 0.5 cm 1. The equilibrium vibrational constants of e 1524.80(50) cm 1 and 1 were obtained for the A 6 state use x e 22.28(25) cm ing the experimental T 0 and T 1 values and the calculated T 2 value. The rotational constants of Table I provide the equilibrium rotational constants of B e 5.342 675(48) cm 1 and 1 for the A 6 state. An experimene 0.141 349(88) cm tal bond length of r e 1.786 485 0(80) has been obtained for the A 6 state using the equilibrium rotational constant. The ground state equilibrium constants were also determined using the constants for the v 0, 1, and 2 vibrational levels Table I . A complete set of the equilibrium constants for the ground and excited state is provided in Table II. It is interesting to compare our experimental values to those obtained in our theoretical calculations. The experimental ground state values of e 1656.051 15(54) cm 1, 1 e x e 30.491 23(26) cm , and r e 1.655 411 1(57) are in excellent agreement with the theoretical values of e 1653.6 cm 1, e x e 31.0 cm 1, and r e 1.654 obtained using the corrected potential as described in detail in the text theoretical part . Similarly the excited state experimental values of e 1524.80(50) cm 1, e x e 22.28(25) cm 1,

Downloaded 20 Aug 2001 to 129.97.47.16. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp

J. Chem. Phys., Vol. 115, No. 3, 15 July 2001 TABLE I. Rotational constants in cm X Constantsa Tv Bv D v 104 H v 109 L v 1012
v

The A
1

-X

transition of CrH

1315

for the X

and A

states of CrH. A
6

0.0 6.131 737 4 18 3.489 22 24 12.471 27 0.050 362 9 64 0.029 03(62)

1595.068 70 14 5.950 759 3 96 3.440 73 88

3129.154 95 48 5.769 783 84 3.385 24

Dv Hv Lv
v

104 107 1011

0.046 454 14 0.0205b

0.042 577 29 0.0755b

104 103 v 4 D v 10 103 sv 6 sD v 10

Dv a b

0.232 820 6 56 0.0966(54) 7.85(30) 0.005 27 89

0.224 003 49 0.161 31

0.215 132 98 0.268 60

11 552.684 16 75 5.272 000 18 2.5458 12 4.77 22 2.18 14 1.302 076 70 1.5431(46) 0.800 20 2.45(15) 1.575 53 22 8.360(32) 2.98 13 1.017(20) 0.9182 96 6.380 71

13 032.9233 39 5.130 651 86 1.6269 43

1.429 85 20 4.303(20) 9.49(15) 1.5463 12 4.28 11

3.962 22

Numbers in parentheses are one standard deviation in the last digits. Fixed value, see text for details.

and r e 1.786 485 0(80) are also in excellent agreement with our theoretical values of e 1525.2 cm 1, ex e 23.0 cm 1, and r e 1.765 . Signicantly different values of e 1515 cm 1 and e x e 21.5 cm 1 were obtained by Kleman and Uhler5 for the A 6 state from the analysis of their lower resolution spectra.
V. COMPUTATIONAL RESULTS AND DISCUSSION

The X 6 state is derived mostly from the 7 S(3d 5 4s 1 ) ground state of Cr, which has formed a 4s-1s bond with hydrogen. The rst excited state of Cr 5 S(3d 5 4s 1 ) also yields a 6 state, but it is not expected to be strongly bound. However, the second excited state of Cr, 5 D(3d 4 4s 2 ), can bond to hydrogen by 4s4p hybridization, state. In truth, the X and A which gives rise to the A 6 states are somewhat mixed in character, with the X state being derived mostly from 7 S but with some 5 D mixed in and the A state from mostly Cr 5 D. Since the states are derived from Cr occupations with a different number of d electrons, high levels of theory are required to obtain an equivalent description of both states. A slight further complication is curve crossing at large r for the A state due to the Cr 5 S state lying below the 5 D. We rst consider the results obtained by our approaches for the X state, which are summarized in Table III. The
TABLE II. Equilibrium constants in cm of CrH. Constants G 1/2 Be
e 1

MCSCF yields an r e that is too long and e and D 0 values that are too small. Adding more extensive correlation dramatically improves the results, but the IC-MRCI/ANO and IC-MRCI Q/ANO results are only in reasonable agreement with experiment. Using the GEN basis set yields essentially the same results as the ANO set, while expanding the basis set, i.e., ANO , makes only a small change. The MRCI and IC-MRCI results are similar, showing that internal contraction is not the cause for the disagreement with experiment. Increasing the active space, the p or dd approaches, makes only a small improvement. Adding Cr 3s3p correlation makes a bigger change than improving the active space, but the r e is still too long and e is still too small. Accounting for the scalar relativistic effects, using the DK approach, has a signicant effect on the computed spectroscopic con-

TABLE III. The X state spectroscopic constants as a function of level of theory. Method MCSCF IC-MRCI IC-MRCI Q IC-MRCI IC-MRCI Q IC-MRCI dd IC-MRCI Q dd IC-MRCI p IC-MRCI Q p IC-MRCI IC-MRCI Q IC-MRCI 3s3p IC-MRCI Q 3s3p MRCI MRCI Q DK-MRCI DK-MRCI Q Experiment
a b

Basis ANO ANO ANO ANO ANO ANO ANO ANO ANO GEN GEN GEN GEN GEN GEN GEN GEN

r e () 1.721 1.684 1.685 1.681 1.682 1.679 1.679 1.683 1.683 1.683 1.684 1.673 1.670 1.681 1.682 1.669 1.669 1.655a

e (cm

D 0 (eV) 1.39 2.06 2.16 2.09 2.18 2.08 2.13 2.06 2.16 2.04 2.16 1.98 2.16 2.07 2.17 2.15 2.17 1.93 0.07b

for the X

and A

states
6

A 14 10 18 57 54 26

re
e ex e a

1595.068 70 6.222 226 42 0.180 978 08 1.655 411 1 1656.051 15 30.491 23

1480.2391 40 5.342 675 48 0.141 349 88 1.786 485 0 80 1524.80 50 a 22.28 25 b

1525 1581 1577 1586 1584 1593 1596 1586 1582 1581 1577 1595 1601 1585 1581 1631 1628 1656a

Numbers in parentheses are one standard deviation in the last digits. See text for details.

This work. Reference 27.

Downloaded 20 Aug 2001 to 129.97.47.16. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp

1316

J. Chem. Phys., Vol. 115, No. 3, 15 July 2001 TABLE IV. The A state spectroscopic constants as a function of level of theory. A state 1 )

Bauschlicher et al.

Method MCSCF IC-MRCI IC-MRCI Q IC-MRCI IC-MRCI Q IC-MRCI dd IC-MRCI Q dd IC-MRCI p IC-MRCI Q p IC-MRCI p dd IC-MRCI IC-MRCI Q IC-MRCI 3s3p IC-MRCI Q 3s3p MRCI MRCI Q DK-MRCI DK-MRCI Q Experiment
a b

Basis ANO ANO ANO ANO ANO ANO ANO ANO ANO ANO GEN GEN GEN GEN GEN GEN GEN GEN

r e () 1.876 1.782 1.774 1.779 1.771 1.776 1.774 1.781 1.773 1.780 1.772 1.778 1.764 1.779 1.771 1.783 1.775 1.786b

e (cm

T e (cm 1) 8126 10 611 11 088 10 517 10 993 10 796 10 925 10 628 11 137 10 513 10 990 10 513 11 270 10 478 10 946 10 362 10 758 11616b

TMa 0.7101 0.7194 0.7149 0.6986 0.6943 0.7183 0.7205 0.7339 0.7308

1337 1516 1537 1521 1542 1525 1531 1520 1538 1519 1539 1506 1514 1515 1535 1494 1517 1525b

The dipole transition moment, in a.u., at 1.69 . Present work.

stants, and we obtain our best agreement with experiment. However, even at this level, the r e is still too long and e is still too small. While the calculation of D 0 is not the main focus of this work, we note that essentially all of our computed MRCI values are larger than experiment.27 We suspect that the Q values are more accurate than those without the correction, and the Q values suggest a D 0 value of about 2.15 eV. The smallest value computed is 1.98 eV at the IC-MRCI 3s3p level, but this approach has a large Q effect, which is consistent with correlating 15 electrons. We therefore believe that it is better to take the IC-MRCI as our lower bound of 2.06 eV, leading to our best theoretical estimate of 2.15 0.1 eV for the D 0 of CrH. In Table IV we summarize the results obtained for the A state. As expected the MCSCF results do not agree well with experiment, and more extensive correlation must be added. While the IC-MRCI r e is in good agreement with experiment, the e and T e values are too small. The IC-MRCI Q improves T e , but degrades r e and e . The effect of the Q correction is much larger than for the X state. The ANO, GEN, and ANO results are sufciently similar so that basis set incompleteness does not appear to be a problem, nor does internal contraction, as measured by the difference between the IC-MRCI and MRCI results using the GEN basis. Adding the d-d correlation to the active space reduces the size of the Q effect and the e is now in better agreement with experiment. Adding 3s3p correlation in the IC-MRCI 3s3p calculation yields e and T e values that are too small. The Q improves e and T e values, but results in an r e value that is too small. The DK-MRCI Q gives reasonable agreement with experiment for r e and e , but its T e value is too small. We have computed the A-X transition moment at several r values for most of the methods considered in this work.

The moment is about 1.3 a.u. at 1 and decreases in a nearly linear manner until about 2.5 . Due to the avoided crossing in the A state, the moment changes dramatically near 3 see Fig. 2 , and this change depends strongly on the level of theory. Luckily, the region of the change in the transition moment is not in the FranckCondon region for the v less than or equal to 3, and therefore it has only a very small affect on the reported A values. The change in the

FIG. 2. The computed A-X transition moment for three levels of theory and the corrected transition moment see text .

Downloaded 20 Aug 2001 to 129.97.47.16. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp

J. Chem. Phys., Vol. 115, No. 3, 15 July 2001

The A

-X

transition of CrH

1317

TABLE V. Summary of the computed spectroscopic constants using full potentials. X state D 0 (eV) 2.06 2.16 2.06 2.16 2.08 2.13 2.15 2.17 1.98 2.16 2.17 1.93 0.07c
e (cm

Method/Basis ICMRCI/ANO ICMRCI Q/ANO ICMRCI p /ANO ICMRCI Q p /ANO ICMRCI dd /ANO ICMRCI Q dd /ANO DK-MRCI/GEN DK-MRCI Q/GEN ICMRCI 3s3p/GEN ICMRCI Q 3s3p/GEN Correcteda Experiment Method/Basis ICMRCI/ANO ICMRCI Q/ANO ICMRCI p /ANO ICMRCI Q p /ANO ICMRCI dd /ANO ICMRCI Q dd /ANO DK-MRCI/GEN DK-MRCI Q/GEN ICMRCI 3s3p/GEN ICMRCI Q 3s3p/GEN Correcteda Experimentb
a

r e () 1.684 1.685 1.683 1.683 1.679 1.679 1.669 1.669 1.673 1.670 1.654 1.655b r e () 1.782 1.774 1.781 1.773 1.776 1.774 1.783 1.775 1.778 1.764 1.765 1.7865

e (cm

e x e (cm

1580.3 1575.8 1585.2 1580.2 1592.6 1595.7 1631.1 1628.4 1595.2 1601.3 1653.8 1656.05b
1

30.2 29.6 29.9 29.5 30.3 30.0 30.7 30.6 30.9 30.1 31.0 30.49b A(s 1) 0.111( 0.116( 0.108( 0.113( 0.116( 0.117( 0.112( 0.117( 0.112( 0.120( 0.120( 7) 7) 7) 7) 7) 7) 7) 7) 7) 7) 7) Error T 0 (cm 1) 972.3 483.5 955.8 435.7 789.3 658.5 1256.8 849.0 1081.9 310.2 656.8

A state 1 )

e x e (cm

1516.8 1537.4 1520.5 1538.1 1525.7 1531.8 1494.5 1517.5 1506.4 1544.6 1525.2 1524.80

22.8 23.7 22.3 23.3 23.6 24.0 22.8 23.8 21.8 22.9 23.0 22.28

See text for denition. Present work. c Reference 27.

b

transition moment with level of theory at 1.69 is representative of the linear region and therefore we give the moment at this r value in Table IV. While it is too expensive to perform IC-MRCI p dd calculations at many r values, it is possible to compute the transition moment at this level at 1.69 . We rst note that the computed transition moments are not strongly affected by the different levels of theory. It is interesting that the inclusion of the Cr 4p orbitals in the active space decreases the moment somewhat, which is supported by the IC-MRCI-p dd calculation. The calculations using the GEN and ANO basis sets suggest that basis set improvements will increase the moment. The difference between the MRCI and IC-MRCI calculations show that eliminating the internal contraction increases the computed transition moment slightly. We conclude that our computed transition moments are accurate because the values vary only slightly with level of theory. On the basis of the calibration calculations, it appears that one must account for both 3s3 p correlation and scalar relativistic effects. The results also suggest that using a very large active space would be ideal. Unfortunately we are unable to include all of these effects in a calculation for which it is practical to compute an entire surface. We have therefore computed full potentials for a subset of the methods considered in Tables III and IV, and the results are given in Table V. Following Partridge and Schwenke,28 we tried merging these potentials into one that accounts for all of these effects.

Our best derived potential corrects the DK-MRCI Q potential for the 3s3 p correlation, i.e., E corrected E DK-MRCI Q 3s3 p E ICMRCI Q

E IC-MRCI Q .

For the X state, this potential yields excellent agreement with experiment, excluding the somewhat uncertain D 0 value. The same correction for the A state yields very good results for e and e x e , but the r e value is a bit short, see Table V. We should note that the difference between these A state results and the older values5 was part of the motivation for the new experimental analysis of the A state. While the corrected potential did detect some problem with the older spectroscopic constants, it is far from perfect, the A state r e is too short and T 0 is too small. We derive a corrected transition moment in a manner analogous to the potentials, namely TM corrected TM DK-MRCI TM IC-MRCI . This moment has a small bump at about 2.6 and becomes negative at CrH distances larger than 2.8 , see Fig. 2. Since these features may be nonphysical, we also test three modied versions: 1 where the negative values are removed by setting the transition moment to zero at distances TM ICMRCI 3s3p

Downloaded 20 Aug 2001 to 129.97.47.16. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp

1318

J. Chem. Phys., Vol. 115, No. 3, 15 July 2001

Bauschlicher et al.

TABLE VI. Einstein coefcients in s 1 for A-X transition and A state radiative lifetimes computed using the corrected potentials. Av
v

Value 0.120( 0.134( 0.102( 0.624( 0.584( 0.192( 0.180( 0.242( 0.730( 7) 6) 4) 6) 6) 6) 4) 6) 6) s 0.75 0.71

Av

Value 0.270( 0.211( 0.198( 0.852( 0.445( 0.640( 0.116( 0.211( 0.146( 6) 6) 4) 5) 6) 6) 6) 6) 4) s 0.69 0.67

A 00 A 01 A 02 A 10 A 11 A 12 A 13 A 20 A 21 Radiative lifetimes v 0 1

A 22 A 23 A 24 A 30 A 31 A 32 A 33 A 34 A 35
v 2 3

scription of the states. Scalar relativistic effects are found to be important. Cr 3s3 p correlation is also important. The computed Einstein A values and radiative lifetimes do not vary strongly with level of theory, leading us to conclude that our computed values are accurate. The calculations also suggest that the D 0 of CrH is slightly larger than the most recent experimental value.
ACKNOWLEDGMENTS

The National Solar Observatory is operated by the Association of Universities for Research in Astronomy, Inc., under contract with the National Science Foundation. This research was supported by the NASA laboratory astrophysics program and the Natural Sciences and Engineering Research Council of Canada NSERC .
A. G. Gaydon and R. W. B. Pearse, Nature London 140, 110 1937 . S. OConnor, J. Phys. B 2, 541 1969 . 3 R. E. Smith, Proc. R. Soc. London, Ser. A 332, 113 1973 . 4 B. Kleman and B. Liljeqvist, Ark. Fys. 9, 345 1955 . 5 B. Kleman and U. Uhler, Can. J. Phys. 37, 537 1959 . 6 S. OConnor, Proc. R. Ir. Acad., Sect. A, Math. Astron. Phys. Sci. 9, 95 1967 . 7 R. S. Ram, C. N. Jarman, and P. F. Bernath, J. Mol. Spectrosc. 161, 445 1993 . 8 R. S. Ram and P. F. Bernath, J. Mol. Spectrosc. 172, 91 1995 . 9 D. Dai and K. Balasubramanian, J. Mol. Spectrosc. 161, 455 1993 . 10 O. Engvold, H. Wohl, and J. W. Brault, Astron. Astrophys., Suppl. Ser. 42, 209 1980 . 11 B. Lindgren and G. S. Olofsson, Astron. Astrophys. 84, 300 1980 . 12 J. D. Kirkpatrick, F. Allard, T. Bida, B. Zuckerman, E. E. Becklin, C. Chabrier, and I. Baraffe, Astrophys. J. 519, 834 1999 . 13 S. M. Corkery, J. M. Brown, S. P. Beaton, and K. M. Evenson, J. Mol. Spectrosc. 149, 257 1991 . 14 K. Lipus, E. Bachem, and W. Urban, Mol. Phys. 73, 1041 1991 . 15 B. A. Palmer and R. Engleman, Atlas of the Thorium Spectrum Los Alamos National Laboratory, Los Alamos, 1983 . 16 H.-J. Werner and P. J. Knowles, J. Chem. Phys. 89, 5803 1988 ; P. J. Knowles and H.-J. Werner, Chem. Phys. Lett. 145, 514 1988 . 17 B. A. Hess, Phys. Rev. A 33, 3742 1986 . 18 C. W. Bauschlicher, Theor. Chim. Acta 92, 183 1995 . 19 T. H. Dunning, J. Chem. Phys. 90, 1007 1989 . 20 R. A. Kendall, T. H. Dunning, and R. J. Harrison, J. Chem. Phys. 96, 6796 1992 . 21 MOLPRO is a package of ab initio programs written by H.-J. Werner and P. J. Knowles, with contributions from J. Almlof, R. D. Amos, A. Berning et al., The closed shell CCSD program is described in C. Hampel, K. Peterson, and H. J. Werner, Chem. Phys. Lett. 190, 1 1992 . 22 MOLECULE-SWEDEN is an electronic structure program written by J. Almlof, C. W. Bauschlicher, M. R. A. Blomberg, D. P. Chong, A. Heiberg, S. R. Langhoff, P.-. Malmqvist, A. P. Rendell, B. O. Roos, P. E. M. Siegbahn, and P. R. Taylor. 23 See AIP Document No. E-JCPSA6-115-034125 for a table containing the line list. This document may be retrieved via the EPAPS homepage http:// www.aip.org/pubservs/epaps.html or from ftp.aip.org in the directory /epaps/. See the EPAPS homepage for more information. 24 J. M. Brown and D. J. Milton, Mol. Phys. 31, 409 1978 . 25 R. M. Gordon and A. J. Merer, Can. J. Phys. 56, 642 1980 . 26 J. M. Brown private communication , see Ref. 7. 27 Y.-M. Chen, D. E. Clemmer, and P. B. Armentrout, J. Chem. Phys. 98, 4929 1993 . 28 H. Partridge and D. W. Schwenke, J. Chem. Phys. 106, 4618 1997 .
2 1

larger than 2.8 , 2 where the bump and negative regions are removed by connecting, with a spline function, the computed moment at r(CrH) 2.4 with a value of zero at r(CrH) 4.5 , and 3 where the bump and negative regions are removed by connecting the computed moment at r(CrH) 2.4 with a value of zero at r(CrH) 2.8 . For v 3 and v 5, the A values are essentially independent of these modications to the transition moment; our reported A values are computed using the transition moment where the value is set to zero for r(CrH) greater than 2.8 . In Table V we report A 00 values obtained using the computed and corrected potentials. One observation is that the A 00 values do not vary signicantly with level of theory, ranging from 0.108( 7) to 0.120( 7). We therefore conclude that our computed value is probably accurate to 10% 20%. In Table VI we report additional A values and some radiative lifetimes for the corrected potentials. The small difference in the r e values for the X and A states results in a few large A values for each v level. The radiative lifetimes vary by even less with level of theory than A 00 , and so we expect these to be even more accurate. The radiative lifetimes vary slightly with v .
VI. CONCLUSIONS

We have reinvestigated the A 6 -X 6 transition in CrH. The 1-0 and 1-1 bands have been measured and improved spectroscopic constants have been determined. A value for the 2-0 band origin has been obtained from the band head using estimated spectroscopic constants. These data provide a set of much improved equilibrium vibrational state. and rotational constants for the A 6 We have studied the X and A states of CrH and found that high levels of theory are required to obtain a good de-

Downloaded 20 Aug 2001 to 129.97.47.16. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp